Yes the start/end row keys are optional. The "usual" way to use them refers to the lexicographical nature of the row keys. For example you'd want to start scanning at row key "aaa" and you want all the stuff that's "aaa*" but you want to stop at "aab". Also if the row keys are dates, you are able to scan a whole day in the same way (grabbing everything in the hours and minutes of that day).
J-D On Sat, Jul 25, 2009 at 4:08 PM, Saptarshi Guha<[email protected]> wrote: > Ah, thank you. When creating the Scan, i though I had to specify a > start key, but what happens if I dont know the min key? However I > misread the API and I can create a scan if I instantiate the class > without a start key. > > Thank you. > Saptarshi > > I'll read things a bit more carefully next time around before posting. > > On Sat, Jul 25, 2009 at 1:00 PM, Jean-Daniel Cryans<[email protected]> > wrote: >> If you have only 1 region, which I'm sure you do unless your rows are >> 100MB big, that region doesn't have a start nor an end row key. The >> web UI can show it to you if you click on your table name. >> >> What were you trying to do with the table start and end row key? Maybe >> there's another solution to your problem we can discuss. >> >> J-D >> >> On Sat, Jul 25, 2009 at 3:55 PM, Saptarshi Guha<[email protected]> >> wrote: >>> Hello, >>> I tried creating a table with about 20 row keys. Now in order to find >>> the first row key, I did >>> >>> byte[][] start = tz.getStartKeys(); >>> byte[][] ends = tz.getEndKeys(); >>> >>> But the length of start is 0 and same for ends. >>> >>> I wanted to confirm that is an issue on my end, or is there something >>> up with the function. >>> >>> Thank you >>> Saptarshi >>> >> >
